SUSY morph studies of inclusive spectra 04.09.2009 Max Baak and - PowerPoint PPT Presentation

SUSY morph studies of inclusive spectra 04.09.2009 Max Baak and Stefan Gadatsch

Test of morph- & fit stability: M effective � FourMeff, m 0 =1140 fixed � kernel p.d.f. with mirroring turned on for both sides � morph samples with m 1/2 =570, 690, 750 � fit data at m 1/2 =630 � generate and fit 1000 toy samples, each 5000 events with morphed p.d.f.

Test of morph & fit stability: M transverse � TransMass , m 0 =1140 fixed � kernel p.d.f. with mirroring turned on for both sides � morph samples with m 1/2 =570, 690, 750 � fit data at m 1/2 =630 � generate and fit 1000 toy samples, each 5000 events with morphed p.d.f.

Test of mSUGRA grid continuity: M effective � FourMeff , m 0 =1140 fixed � kernel p.d.f. with mirroring turned on for both sides � morph samples with m 1/2 =570, 690, 750 � fit data at m 1/2 =630 � generate 1000 toy samples, each 5000 events from real p.d.f. at m 1/2 =630 � fit morphed p.d.f. to the toydata

Test of mSUGRA grid continuity: M transverse � TransMass , m 0 =1140 fixed � kernel p.d.f. with mirroring turned on for both sides � morph samples with m 1/2 =570, 690, 750 � fit data at m 1/2 =630 � generate 1000 toy samples, each 5000 events from real p.d.f. at m 1/2 =630 � fit morphed p.d.f. to the toydata

Current problems in RooFit � problems with p.d.f. normalization in RooFit � for every value of a i the normalization of the stretched p i is recalculated • the calculation is not factorized as should be done • started to implement analytical integral ourselves � normalization for two observables M effective and M transverse is too slow to be practical � all the plots shown here have been made with keyspdf � ideally for many templates want to work with histograms which are computionally fast � when we use histograms the numerical integration is a nightmare • discontinuous jumps cannot be modelled with a spline � meeting Wouter today to discuss these problems